X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.h;h=7c08fcc78ec39faa4f09396260a81f9c67320c21;hb=0f2d33d4ee4291d61750769c2918263c7e485ffa;hp=c0dab2b6155ab6db9af1127c2b2416893849c61d;hpb=143f18091facab1154eb750b6ed530905bfed889;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.h b/XMLTree.h
index c0dab2b..7c08fcc 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -25,6 +25,13 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <cstring>
+
+//KIM : OJO
+//clash between TextCollection/Tools.h and libcds/includes/basics.h
+#undef W
+#undef WW
+#undef Wminusone
+
 #include "bp.h"
 #include <static_bitsequence.h>
 #include <alphabet_mapper.h>
@@ -37,6 +44,8 @@ using SXSI::TextCollection;
 
 #define NULLT -1
 
+#define PERM_SAMPLE 10
+
         // sets bit p in e
 #define bitset(e,p) ((e)[(p)/W] |= (1<<((p)%W)))
         // cleans bit p in e
@@ -53,6 +62,13 @@ typedef struct {
 } range;
 
 
+//KIM : OJO
+// I know this class implements the working draft that we have but the logics seem flawed here...
+// We should have two classes. One XMLTreeBuilder and one XMLTree.
+// XMLTreeBuilder would have OpenDocument, NewOpenTag,... and CloseDocument would return an XMLTree
+// XMLTree would have only an initialized structure. If find it really ugly to check (!finished) or (!initialized)
+// in every function (FirstChild....).
+
 class XMLTree {
    /** Balanced parentheses representation of the tree */
    bp *Par;
@@ -67,7 +83,7 @@ class XMLTree {
    static_bitsequence_rrr02 *EBVector;  
 		      
    /** Tag sequence represented with a data structure for rank and select */
-   static_sequence_wvtree *Tags;
+   static_sequence *Tags;
 
    /** The texts in the XML document */
    TextCollection *Text;
@@ -85,8 +101,21 @@ class XMLTree {
    pb *par_aux;
    TagType *tags_aux;
    int npar;
+   int parArraySize;
    int ntagnames;
    unsigned int *empty_texts_aux;
+
+   // KIM : OJO
+   // I added those two. The TagName array should always contains two special tags
+   // <@> for attributes and <$> for PCDATA.
+   // <$> can never be in a document (since we handle the text differently)
+   // but <@> can be returned by the parser. This boolean is needed for the construction
+   // of the Tag bitmap to know if <@> must be taken into account or not
+   bool found_attributes;
+
+   // KIM : OJO
+   // Allows to disable the TextCollection for benchmarkin purposes
+   bool disable_tc;
    
 public:
 
@@ -206,14 +235,17 @@ public:
    /** ParentNode(d): returns the parent node of document identifier d. */
    treeNode ParentNode(DocID d);
 
-   /** OpenDocument(empty_texts,sample_rate_text): initilizes the construction
+   /** OpenDocument(empty_texts,sample_rate_text,dtc): initilizes the construction
     * of the data structure for the XML document. Parameter empty_texts 
     * indicates whether we index empty texts in document or not. Parameter 
     * sample_rate_text indicates the sampling rate for the text searching data
     * structures (small values get faster searching but a bigger space 
-    * requirement). Returns a non-zero value upon success, NULLT in case of 
+    * requirement). dtc disable the use of the TextCollection
+    * (i.e. everything is considered an empty text *)
+    * Returns a non-zero value upon success, NULLT in case of 
     * error. */
-   int OpenDocument(bool empty_texts, int sample_rate_text);
+
+   int OpenDocument(bool empty_texts, int sample_rate_text, bool dtc);
 
    /** CloseDocument(): finishes the construction of the data structure for 
     * the XML document. Tree and tags are represented in the final form, 
@@ -251,6 +283,25 @@ public:
     * Returns NULL in case that the tag identifier is not valid.*/
    unsigned char *GetTagName(TagType tagid);
 
+
+   // OJO
+   /** GetTagName(tagid): returns the tag name of a given tag identifier.     
+    *  The result is just a reference and should not be freed by the caller.
+    */
+   const unsigned char *GetTagNameByRef(TagType tagid);
+
+   //OJO
+   /** RegisterTag adds a new tag to the tag collection this is needed
+    * if the query contains a tag which is not in the document, we need
+    * to give this new tag a fresh id and store it somewhere. A logical
+    * choice is here.
+    * We might want to use a hashtable instead of an array though.
+    */
+   TagType RegisterTag(unsigned char *tagname);
+
+   bool EmptyText(DocID i) {
+      return Text->EmptyText(i);
+   }
    /** Prefix(s): search for texts prefixed by string s. */
    TextCollection::document_result Prefix(uchar const *s) {
       return Text->Prefix(s);
@@ -333,9 +384,12 @@ public:
    /** GetText(d): returns the text corresponding to document with
     * id d. */
    uchar* GetText(DocID d) {
-      Text->GetText(d);
+      return Text->GetText(d);
    }
    
+   TextCollection *getTextCollection() {
+      return Text;
+   }
    /** Save: saves XML tree data structure to file. */
    void Save(unsigned char *filename);